Multi-expression programming based prediction of the seismic capacity of reinforced concrete rectangular columns

Abstract

This article presents an innovative artificial intelligence based multi-expression programming approach to predict the seismic capacity of reinforced concrete rectangular columns. For the evolution of these multi-expression programming based computational models, an investigational test database created by pacific earthquake engineering research centre was employed. This database consisted of the experimental record of 250 specimens of reinforced concrete rectangular columns exposed to seismic loading. The overall seismic capacity of the aforementioned columns was considered to be the function of their flexural and shear capacities, for the prediction of which five input variables were used. Once these models were evolved utilizing MEPX version 2022.3.19.0-beta, their performance was determined on the basis of ten most frequently used performance indicators, one of which is the coefficient of determination i.e., R2. Based on the findings of this research, a strong correlation (i.e., an R2 of 0.9747 and 0.9664 for the flexural and shear capacity prediction models respectively) was found to exist between the predicted and tested seismic capacities of reinforced concrete rectangular columns. Moreover, the proposed seismic capacity prediction models were found to outperform that of ACI 318-19 in terms of precision and prediction accuracy. The proposed seismic capacity prediction models were also found to be well trained on the correlation between input and output variables which verifies their ability to capture the underlying physical processes involved in the seismic action on reinforced concrete rectangular columns. Therefore, the proposed multi-expression programming based computational models can be confidently recommended for predicting the seismic capacity of reinforced concrete rectangular columns for practical design applications.